External beam radiation therapy is a major and effective treatment modality for prostate cancer. Despite the significant improvement over the past decades, tumor control rates and radiation induced complication rates have been suboptimal. One major cause is the geometrical and anatomical variations during the treatment course, such as patient setup and organ motion at each step of treatment process. To account for these uncertainties, margins are added during treatment planning to ensure adequate target coverage. The margins increase the irradiated volume which may encompass a significant portion of critical organs. To reduce margins safely, accurate information of patient anatomy under the treatment condition are required. Recent advancement in delivery system includes mounting a kV X-ray imaging system capable of cone-beam CT on board the linear accelerator. High spatial resolution CT images from the system have good soft tissue contrast to localize prostate gland accurately at treatment condition. However, the benefits of cone-beam CT in reducing uncertainties in clinical environment are unknown and no dosimetric evaluation has been performed in determining the optimal margin reduction using cone-beam CT. The hypothesis of this project is that, optimal and efficient use of the on-board imaging system, integrated with offline adaptive plan optimization and online treatment intervention, can significantly increase the dose gradient between the tumors and critical organs. Three specific aims will be pursued. (1) Develop image guidance strategies for prostate cancer treatment that include online geometric treatment intervention (plan correction) and offline adaptive dosimetric plan re-optimization. The goal of online process is to eliminate uncertainties from setup and inter-fraction motion. The residues and intra-fraction motions are incorporated in the offline process. (2) Evaluate and optimize image guidance strategies through comparative planning studies against the current standard methods and validate cone-beam CT for image guidance. (3) Perform pre-clinical studies to test feasibility and efficiency of the image guidance strategies. The long-term goal of this project is to implement a clinical process of image guided radiotherapy of prostate cancer, in which geometric variations will be incorporated into treatment planning and delivery through image feedback. The adoption of image guidance will allow significantly increased local control and reduced radiation induced complications in prostate cancer patients.